🧠 Neural Networks for Character Recognition

Author: Kaitlyn Kirt
Course: CMOR 220
Term: Spring 2024
Project: Neural Networks - Pattern Recognition
Last Modified: April 26, 2024

---

📋 Overview

This project builds a basic neural network from scratch using NumPy to recognize letters represented as binary vectors. It simulates recognizing noisy versions of the letters R, I, C, and E using a feedforward neural network with sigmoid activations.

---

📂 Files

• Project11.ipynb: The main Jupyter Notebook implementing all logic for training and testing the neural network.
• README.md: Description and instructions for running the project.

---

🔍 Techniques Used

• 🧮 Feedforward Neural Network with:
  • One hidden layer
  • Manual backpropagation
• 🎯 Binary classification using 2-bit target outputs
• 🧪 Repeated testing with noisy letter variants via random bit flips
• 📊 Visualization of the letter patterns as 5x5 binary matrices

---

🛠️ Requirements

pip install numpy matplotlib